10:16 <russell--> KanjiMonster: i added some debugging printk's to the 4.14.113 + revert to see why it didn't warm-boot hang, and it was because the _old macro wasn't actually setting the value, so, for example the problematic command, shows:

10:16 <russell--> [ 0.120068] getCx86(c2)=08

10:16 <russell--> [ 0.130000] setCx86(c2,88)

10:16 <russell--> [ 0.130000] getCx86(c2)=08

10:16 <russell--> (i read the value before and after the set)

10:17 <KanjiMonster> russell--: yes, that is what https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=f25f64ed5bd3c2932493681bdfdb483ea707da0a also said (when this was done the first time)

10:18 <KanjiMonster> I wonder if this is a bios issue

10:20 <russell--> if it is a bios issue, it probably isn't fixable

10:20 <russell--> i mean, we can't fix the bios

10:21 <russell--> i haven't grasped what is fiddling with the SUSP# line

10:28 <russell--> there are a grand total of 6 calls to the setCx86_old macro during initialization

10:38 <russell--> i guess the outstanding question is, what is affected (if anything) beyond SC1100 or maybe even just the soekris net48xx devices, and how to restrict the change to not enable SUSP# to just those devices affected

11:25 <KanjiMonster> russell--: the pdf says to toggle gpio63 to clear the susp# signal, so that should be doable by the kernel before enabling it. no idea how to implement it though

15:29 <russell--> svanheule: regarding gs1900, i'm i understanding i can flash an initramfs image and sysupgrade to the rearranged partitioning from that?

15:54 <SlimeyX> what os is being used these days to build openwrt and u-boot

16:00 <schmars[m]> i know debian fedora and alpine work fine

16:02 <SlimeyX> heh arch is giving me odd fits

16:04 <SlimeyX> going to revisit these bsap units again now that things have calmed down

17:27 <svanheule> russell--: yes, if you manage to build an initramfs image, you can (force) flash that. There's a size check in the build for safety

20:59 <russell--> SlimeyX: i use arch

21:00 <russell--> i haven't noticed any problems recently

22:17 <Habbie> ugh, system header leakage

22:17 <Habbie> any general hints for fixing that?

22:17 <Habbie> In file included from /usr/include/wchar.h:30,

22:17 <Habbie> from /home/peter/openwrt/staging_dir/toolchain-arm_arm926ej-s_gcc-13.3.0_musl_eabi/arm-openwrt-linux-muslgnueabi/include/c++/13.3.0/cwchar:44,

22:17 <Habbie> etc.

23:04 <Habbie> ok, somehow -I/usr/include made it into my compiler call

23:05 <Habbie> (before the -I that has the right dir for wchar.h)

23:17 <Mangix> Habbie: grep includedir=/ /home/peter/openwrt/staging_dir/target-arm_arm926ej-s_gcc-13.3.0_musl_eabi/usr/lib/pkgconfig

23:18 <Habbie> grep: /home/peter/openwrt/staging_dir/target-arm_arm926ej-s_gcc-13.3.0_musl_eabi/usr/lib/pkgconfig: No such file or directory

23:18 <Habbie> but i found a few deps that this software (voorkant.org) looks for that i did not put into DEPENDS

23:18 <Mangix> i got the directory wrong then

23:18 <Mangix> maybe grep includedir=/ /home/peter/openwrt/staging_dir/target-arm_arm926ej-s_gcc-13.3.0_musl/usr/lib/pkgconfig

23:19 <Habbie> and -r

23:19 <Habbie> ok, that's empty

23:19 <Habbie> but it looks like it was missing deps being picked from host because they were not found in openwrt cross

23:20 <Mangix> ah right

23:20 <Mangix> that is an issue, yes

23:20 <Habbie> i learn every day!

23:20 <Habbie> i have -great- news (really)

23:20 <Habbie> i have reduced my failing build to 'looks like it needs -latomic' :)

23:21 <Mangix> latomic is lovely

23:22 <Habbie> yeah

23:25 <Habbie> quickjs_options.append_compile_args('c', '-latomic')

23:25 <Habbie> compile args .. what was i thinking :D

23:28 <Habbie> (meson, indeed, nicely passes -latomic to every .o generation, which is useless)

23:35 <Habbie> it built!